CSV vs JSON in Magento: Which to pick for 1M SKUs
Which format is faster, easier to edit, friendlier to UTF-8 and produces fewer errors at scale.
Published 2026-10-08 • Category: Data • Reading time: 9 min
Short answer
If you’re exporting from Excel or 1C — CSV. If the data already lives in an API or NoSQL store — JSON. The rest of the article is why, and the trade-offs.
CSV: pros and cons
Pros: native support in Excel, Google Sheets, LibreOffice. Every ERP can export it. Files are compact. Easy to read by eye.
Cons: no schema. A single stray newline inside a description and the file is unreadable. BOM encoding, delimiters, quotes — eternal pain.
JSON: pros and cons
Pros: hierarchical structures (configurable products, variants, multi-store values — natural). Schema via JSON Schema. Easy to parse in any language.
Cons: 2-3× larger than CSV for the same data. Can’t be edited in Excel. One character outside the syntax and it stops being JSON.
Benchmark on 1M SKUs
| Metric | CSV | JSON |
|---|---|---|
| File size | 420 MB | 1.2 GB |
| Disk read time | 3 s | 11 s |
| PHP parse time | 14 s (streaming) | 38 s (chunked JSONL) |
| Total import | 21 min | 34 min |
When JSON still wins
Three scenarios:
- Data already lives in MongoDB / PostgreSQL JSONB
- You have heavy localisation: 20 languages with 50 attributes each
- Webhook integrations where the sender doesn’t ask your opinion
For those — the JSON Datapump plugin for Magmi. It streams a JSONL file and processes it as efficiently as CSV.
Recommendation
Large clients often combine: hourly stock delta — CSV (easy to generate from an ERP), daily full product refresh — JSONL (because all ERP attributes are available).
Leave a Reply